Nonlinear control for vehicle active suspension systems

Abstract

Driving safety and ride comfort problems of an automobile have become the main concern of modern society and the automobile manufacturers since millions of people suffer injuries from road accidents per year. Ride comfort and maneuverabilities of a vehicle are closely related to its suspension system since a well-designed suspension can not only enable the driver to keep authority over the vehicle in critical situations, but also provide a high level of ride comfort to prevent physical fatigue of the driver. Traditional passive suspensions or semi-active ones are inadequate in improving ride comfort or road holding, especially under those extreme poor road conditions. In contrast, active suspension systems possess a significant potential to enhance substantial performance improvements of the ride comfort and vehicle maneuverability. Based on the full understanding of the state of the art in vehicle suspension systems, the presented dissertation focuses on the key issues in nonlinear control design for vehicle active suspensions, particularly in handing actuator-related challenges and suspension model parameter uncertainties, to improve ride comfort and driving safety. The Thesis proposes fve novel nonlinear vehicle active suspension control approaches, which significantly enhance the performances of suspension systems by adjusting the controller design parameters. Thus, ride comfort is improved while the suspension deflection and the dynamic wheel load remain uncritical. The main researches are summarized as follows. The first concept is the finite-time controller structure, which significantly suppresses the unknown disturbance effect while guarantees the transient property and steady-state accuracy of the closed-loop system. The required disturbance compensator of the unknown external disturbances is provided by a finite-time estimator concept based on sliding-mode algorithms. Moreover, the overall controller with the compensator for the closed-loop system is continuous, which provides some distinct advantages for practical applications. The performance of the control concept is successfully validated in experiments on a quarter-car test setup for an active suspension. To overcome the high power demand drawback of an active suspension system, the second control method presents a novel suspension concept called bio-inspired dynamics-based suspension system. It fully exploits and reveals the advantage of benefcial nonlinear stiffness and damping characteristics inspired by the limb motion dynamics of biological systems to realize advantageous nonlinear suspension properties with potentially less energy consumption. The stability of the desired bio-inspired nonlinear dynamics is analyzed employing a common Lyapunov function approach. The potentialof the proposed bio-inspired nonlinear dynamics-based reference model adaptive control method in the energy-efficiency design is illustrated by theoretical analysis and numerical verification, which show that the presented controller achieves performance improvements that are similar to the classical adaptive control method for the active suspension, however, with a lower power demand. Another issue that deserves careful attention is the control method addressing the problems of the parametric uncertainties and non-ideal actuators (dead-zone and hysteresis inputs) in practical active suspension systems, which generally deteriorate the control performance of the suspensions. To remove their effects, an adaptive control algorithm is proposed by constructing a unified framework of non-ideal actuators, and its transient and steady-state performances are guaranteed by a Lyapunov function approach. Moreover, the assumptions on the measurable actuator outputs, the prior knowledge of the actuator parameters and model uncertain parameters are not required in the controller design procedure, which promotes flexibility of the control approach. This adaptive control concept is implemented on a quarter-car setup for employing the suspension performance. As an addition to the actuator nonlinearities mentioned above, the fourth control strategy carries on independently research in the problem of merit attention in actuator saturation nonlinearity with regard to its special properties. When the actuator saturation occurs, extra control force has failed to realize its expected effect, which will also lead to the performance degradation. Thus, considering the infuence of the actuator saturation, a disturbance observer based adaptive control is presented together with addressing the parametric uncertainty and bounded external disturbance for a class of nonlinear system. For improving the precision of control, the modeling inaccuracy and external disturbance are integrated as a lumped disturbance to be estimated by a finite-time disturbance observer to achieve the real-time compensation. Based on tracking error state, the control law is divided into three regions to reduce the effect of the actuator saturation. The effectiveness of the presented approach is illustrated by the application to control a quarter-car active suspension system. The last area of concern is the issue of fault-tolerant control for nonlinear systems. Differing from the existing results, in most of which the effect of actuator failure is neglected in the design and analysis of control systems for simplicity, the thesis considers the generally pure-feedback nonlinear system with multiple actuators for actuation redundancy. In response to prepare for possible actuator failures, the model of actuator failures considered here includes different situations related to a multi-Markovian variable since the modes of the actuator failures are generally random in essence and impossible to anticipate in advance. Furthermore, the nonsmooth dead-zone characteristics in practical actuators are also considered in the actuator models such that the developed actuator model is more general. Thus, the problem of fault-tolerant control for the pure-feedback nonlinear system is investigated by adaptive control method, whose aim is to guarantee that the solution of the closed-loop is a unique and bounded in probability. Moreover, the presented controller can achieve arbitrarily small tracking error in the presence of nonlinear actuators with random failures. Finally, the proposed control algorithm is applied to a quarter-car active suspension and verifed by numerical results to show its practical applicability and effectiveness.